Optimal Usage of Time

**batigol** · 02-12-11, 11:51 AM

Great post man ! Good luck

**Data** · 02-12-11, 12:09 PM

Originally posted by Miz

To those that bet seriously...

I have finished 5 years of backtesting a MLB model I have been working on for decent piece of time. Like most working models it does much better versus openers and the line movement agrees w my model much more than it doesn't. While checking the time stamps on line movements on prior plays that my model identified as bettable, many lines had the value removed within the first 15 minutes they were on the board. Unless you are chained to your computer you will likely miss out on the best lines and your profit will not be optimized. Some services have a board that indicates when lines are posted at various books, but you would still have to be in front of the computer all day to be aware of it, right?

How have others dealt with this obstacle? Some people have told me that they simply stay chained the computer; however, betting is their sole source of income. That isn't the case for me. I work a full time job, and have a family etc. and it just isn't feasible for me. I will likely not bother betting my model if I can't reliably bet into the opening lines. The ROI on closers, just isn't worth the time right now.

Thanks.

I reckon your model uses teams and pitchers and does not look at what particular lineup is going to play. Your dilemma illustrates how far you can get with this approach.

**Peregrine Stoop** · 02-12-11, 12:17 PM

I optimize my time by not even creating a huge PITA model and letting guys like you do it for me

**Miz** · 02-12-11, 01:29 PM

Data, that is actually the next step for me in the model. However, this is where I am right now, as I only have about 4-5 hours a week to look at this stuff. Could it be improved? Yes. Could line up specific inputs improve it, very probably. This is much more like a hobby to me than a way of life. Anyway, even with those inputs, I would think most working models do better against openers and would face the same challenges.

Peregrine, if that is the case, who executes your bets and when?

**That Foreign Guy** · 02-12-11, 07:06 PM

It's allegedly possible to create a bot to make bets for you.

If you learn when books release lines and schedule a scraper to acquire the opening odds (and any other info your model needs) then you don't have to be at your computer to bet in to openers.

**Miz** · 02-12-11, 09:35 PM

Thanks FG. Good point, but a little beyond my programming ability at the moment. Thanks for the genuine feedback though. Cheers

**That Foreign Guy** · 02-13-11, 06:00 AM

There are programs you can buy to automate internet tasks like logging in to sites and submitting forms. They're usually used by spammy internet marketers to jam comments threads etc, but it might be a way to avoid having to program the bot yourself.

**CaptainHindsight** · 02-13-11, 12:37 PM

Some services have a board that indicates when lines are posted at various books, but you would still have to be in front of the computer all day to be aware of it, right?

I would think most books have the same set schedule for when they post their lines. If you identify when these are - why would you have to sit in front of the computer all day? You could potentially have an hour window where you were monitoring openers - esp for baseball.

**Miz** · 02-13-11, 12:53 PM

yeah, that's sort of why I posed the question. It didn't seem all that consistent from day to day from what I saw. I looked at more and more line movements, and quite a few took an hour or more to move, but the vast majority moved within 2 hours... to the point where value was minimized at that point. They really didn't seem to be all that consistent as far as time of day that they opened. Maybe a 2 hour window, and pinny opens at about 4pm-6pm the day before, bookmaker was like 1-3pm, and most of the others wait until about 9pm or so, hoping to use the other books as the litmus test. Maybe some of the time stamps are off too. Not sure.

**Flight** · 02-19-11, 11:27 AM

If the value is removed within 15 minutes, remember that the limits are also down at $250 or so

Tough to get a good ROI on openers, even with higher win %

**Miz** · 02-20-11, 02:55 PM

yep, low limits is a problem too, but I don't have to battle that one just yet. I think they are more like 1000 per game at the open, but still could be a pretty limiting factor at some point

**blackbart** · 02-20-11, 11:45 PM

if you can predict which way the line moves, no need to predict the winner

**batigol** · 02-21-11, 07:37 AM

Very interesting !

**Miz** · 02-21-11, 10:25 AM

Originally posted by blackbart

if you can predict which way the line moves, no need to predict the winner

True, the model output (even with some of its limitations) agrees w the movement and increases as the percent difference increases between my model and the implied winning pct. From 56% at 0, to 75 or 80% at around 20% diff or so. The problem lies in that to maximize profit potential, you need to bet into openers. The ROI is very high versus openers. It drops significantly once the market adjusts (about 2.5% versus closers, with fewer plays available to bet), as other people have models that likely have similar outputs. Consequently, as games appear on the board, about half the time, the line moves in the first 1-2 hours. If you miss the train, you are left betting into a much more efficient line... meaning less potential profit. If you bet at multiple books, the limits for openers isn't a huge problem. At least not for me just yet. But you do have to factor in the time you spend betting, as it is work. It potentially takes time away from other things that I enjoy doing, like exercising, or being with my family etc. I do well enough in my day job, and can't afford to be chained to the computer or constantly checking my phone waiting for lines to open. I figured others had the same time management issues but Data obviously has superior models that exhibit massive returns betting into closers and Peregrine has the model mafia working for him.